安全扫描
OpenClaw
可疑
high confidenceThe skill mostly implements the advertised IMAP → AI → Feishu workflow, but its code contains undeclared credential usage and a token verification call to an external host (geo-api.yk-global.com) that could leak API keys; there are also a few runtime inconsistencies in the code.
评估建议
Key things to consider before installing or running this skill:
- The code will read an OPENAI_API_KEY environment variable (even though metadata says none required). If you provide your primary API key, the skill may send it to https://geo-api.yk-global.com/validate when the key prefix matches certain values — this would expose the key to a third party. Do not set your primary API key in your environment unless you trust that endpoint.
- The skill writes a cache directory (~/.email_assistant_ca...详细分析 ▾
⚠ 用途与能力
The declared purpose (IMAP read, classify, reply suggestions, Feishu push) matches the majority of the code. However the package also contains an unexpected token verification component (verify_token) that POSTs an API key to https://geo-api.yk-global.com/validate and persists verification results to ~/.email_assistant_cache — this external verification behavior is not described in SKILL.md or metadata, and is unrelated to core IMAP/Feishu functions.
⚠ 指令范围
SKILL.md promises only-read behavior and does not mention sending credentials off-host. In practice check_emails.py will: read an OPENAI_API_KEY environment variable (if present), call verify_token which transmits that key to an external verification endpoint for some key prefixes, and write cache files to the user's home directory. The code also includes raw email payloads in the returned email dicts (key 'raw'), which could be forwarded by other code paths. Additionally there are apparent runtime mismatches (e.g., check_emails.py calls self.reply_gen.generate(...) while the reply generator exposes generate_reply(...) — suggesting buggy or inconsistent runtime behavior).
✓ 安装机制
No install spec; it's an instruction-plus-scripts package. Dependencies are typical Python libraries (openai/requests/imapclient/pyyaml). No remote archive downloads or obscure install URLs were found in the provided files.
⚠ 凭证需求
Registry metadata declares no required env vars or primary credential, but the code reads OPENAI_API_KEY from the environment and accepts API keys in config. The verify_token flow will POST an API key (Authorization: Bearer {api_key}) to geo-api.yk-global.com for keys with certain prefixes — this is a disproportionate request for a mailbox-classification skill and is not documented in SKILL.md. The skill also expects IMAP credentials and Feishu webhook/app_token (reasonable for functionality), but these are sensitive and should be documented explicitly.
ℹ 持久化与权限
always: false and no special platform privileges. The skill writes cached verification results to ~/.email_assistant_cache and may create config files as instructed (config/config.yaml). Writing to the home directory is normal for CLI scripts but is persistent and should be noted by users. The skill does not request to modify other skills or system settings.
安全有层次,运行前请审查代码。
运行时依赖
无特殊依赖
版本
latestv1.0.02026/4/21
- Initial release of Email Customer Assistant (邮箱客服助手) for ClawHub and 腾讯Skillhub. - Supports IMAP email reading, AI-powered email classification (urgent, important, normal, can be delayed), and multilingual reply suggestions. - Integrates with Feishu for real-time summary push notifications. - Multiple pricing tiers with varying limits and features; free starter plan available. - Multi-language support: Chinese, English, Japanese, and Korean. - Designed for read-only mailbox access with user confirmation required for all actions.
● Pending
安装命令
点击复制官方npx clawhub@latest install email-intelligence-assistant
镜像加速npx clawhub@latest install email-intelligence-assistant --registry https://cn.longxiaskill.com